Changing Paradigms in Operational Research

Maurice Yolles
Liverpool John Moores University
Cybernetics and Systems, 1998, 29(2)91-112
First submitted: September 5, 1997

Abstract
While paradigms evolve and mature, they may also become bounded through the vary
conceptualisations that originally made them successful. The paradigm of complexity is able
to conceptualise problem situations in terms of certainty, softness, and structure. These
conceptualisations can be used to evaluate how different paradigms change. This can be
illustrated by examining the appearance of new Operational Research paradigms over the
last half century, and examining how they change using landmark theory.

1. The Nature of World Views

This paper is about changes in paradigms, and as an introduction to this notion it is essential
to have an adequate understanding of what it is that we mean by a paradigm. More, paradigms
are world views tht are conceptually close to the idea of weltanschauung, and it is difficult to
find discussions about their relationship in the subject literature. Before entering into the
nature and purposes of this paper, it will thus be useful to first discuss these concepts.

Our base need is to distinguish between two types of world view: informal and formal. We
refer to informal world views as weltanschauungen, introduced into systems by Peter
Checkland in the 1970s. Formal world views are called paradigms, and became important in
the 1970s through the work of Kuhn. Both a distinction and interrelationship between the
ideas of weltanschauung and paradigm has been made by Yolles [1996; 1997] that we shall
briefly consider here.

Human activity can be viewed in a number of different ways. The way in which it is seen by
someone is from a viewpoint that is determined by their beliefs, background, interest, and
environment. It generates a perspective, a mental picture of the relationships and relative
importance of things that is itself a mental model of an activity or situation. This is referred to
as weltanschauung. It may be an individual or a group facility. As a group facility
weltanschauung is seen as a shared world view that occurs through the creation of common
models of reality that are manifested in action. Shared weltanschauung occurs through a
process of socialisation. During this individuals become members of the group when they
assign themselves to it, and identify with it taking on its members’ roles, attitudes,
generalised perspective, or more broadly its norms [Berger and Luckmann, 1964]. Identity is
thus “objectively” defined through the group. However, there is always a distinction between
the individual and the group. The two realities correspond to each other, but are not
coextensive [Ibid. p153].

Now, a paradigm is "the set of views that the members of a...community share" [Kuhn, 1970,
p.176]. The paradigm is more than shared weltanschauung, however. It is shared
weltanschauung together with the explicitly defined propositions that contribute to the
understanding of an observer. When weltanschauungen are formalised they become
paradigms. A formalisation is a language that enables a set of explicit statements to be made
1
about the beliefs and propositions (and their corollaries) that enable everything that must be
expressed to be expressed in a self-consistent way. A formalised non-normative or semi-
formalised shared weltanschauung can be created in the absence of a paradigm, called a
virtual paradigm [Yolles, 1996], that may or may not become a paradigm. Formal
propositions define a logic that establishes a framework of thought and conceptualisation that
enables organised action to occur, and problem situations to be addressed. They also constrain
the way in which situations can be described. Formal logic [Kyburg, 1968, p20] provides a
standard of validity and a means of assessing validity. While groups may offer behaviour in
ways that are consistent with their shared weltanschauung, paradigms emerge when the
groups become coherent through formalisation.

Like weltanschauung, paradigms are belief based, and beliefs are not susceptible to rational
argument. Paradigm stakeholders may thus be unable to release their beliefs easily. While
paradigms can evolve, their degree of evolution is bounded by the capacity of a given belief
system to change. The nature of a paradigm may be explicitly defined as in figure 1. Here, it
is composed of: group culture, attributes of which are attitudes, beliefs, values, and language;
and it has both group norms and (cognitive space) conceptualisation associated with it.

Culture

Attitudes Beliefs Values Language

Cognitive
Normative space
standards Concepts, knowledge &
meaning to construct behaviour
Propositional base.
Action/behaviour Exemplars.
& communication
Paradigm

Figure 1
Context Diagram for a Paradigm

2. Inquiry, Models of Reality, and Paradigms

Situations develop that are perceived to be problematic because we have desired goals or
expected outcomes that do not materialise. It is through an understanding of problem
situations that we are able to pose intervention strategies and take action that deals with the
problems. Understanding derives from the process of inquiry.

The way in which we see a situation and formulate cognitive purposes for an inquiry is
determined by our weltanschauung, and the way in which we formally model it is determined
by our paradigms. In the case that we wish to develop intervention strategies for the situation,
it is through these models that they can be formulated. Checkland describes the need to ensure
that an intervention is sytemically or logically desirable [Patching, 1990, p113]. This arises
because the models of a situation that are intended to represent it (Checkland calls them
relevant systems) are also intended to be relevant to the situation. Resulting strategies of
intervention are “systemically desirable if these ‘relevant systems’ are in fact perceived to be
2
truly relevant” [Checkland and Scholes, 1990, p52]. The question that must be asked is who
determines whether such a model is “truly relevant”, and what criteria do they use? The
criteria will derive from the world views involved, and this includes the paradigm from which
the situation is being modelled, and the weltanschauung of the evaluating person (the who).

Models derive from paradigms that have their own “truths” that generate knowledge. Since
different paradigms are incommensurable, the knowledge that they produce will never be
totally reconcilable across their boundaries. The capacity of a paradigm to describe and
explain “real” situations through its models will be related to its penchant that is responsible
for the generation a specialist type of knowledge, and which implicitly determines cognitive
purposes. While paradigms operate at the level of belief and conceptualisation and generate
cognitive knowledge, cognitive purposes describe the purposes attributable to behaviour in a
given situation, and are commonly expressed in a situation as a method’s mission and associated
goals or aims.

The conceptual explanations that are provided by a model about a situation should be able to
disclose relationships that will be essential to its future stability. If this cannot occur then the
capacity of the paradigm from which the model derives is inadequate. Two things may occur
in this case: (i) either the stakeholders of the paradigm will learn cognitively, and the
paradigm will pass through a change process thereby evolving, or (ii) the paradigm will be
replaced by another that can represent reality more adequately. As an example of this, when a
virtual paradigm is created, if it survives, then it does so by passing through a period of
incremental change until it reaches its maturity.

A new paradigm will provide a new approach to problem situations and pose a different class
of questions through its own set of conceptualisations. “It would pursue its answers with its
own set of essential tools, and often evaluates results according to an evolving set of
standards and challenges. Thus the new paradigm unearths and explains phenomena that
could not have been approached with pre-paradigmatic means. Alternatively, the new
paradigm could be shown to provide better, more compact, and more accurate explanations”
[Guastello, 1997].

A mature paradigm may not have the propositional capacity to “satisfactorily” [Weinberg,
1975, p140] explain a given situation. As a result it will produce models that are incongruent
with perceived behaviour as seen from the perspective of other paradigms, leading to
contradiction (and possibly paradox). As an example of this in physics, two classical theories
developed that attempted to explain the nature of light and how it passes through space
[Hoffman, 1947]. These were the corpuscular and the wave theories, each of which had their
own paradigms. In the corpuscular theory, light was seen as particles, and the properties that
we might assign to them had to satisfy the dynamics of corpuscular bodies. In the wave
theory, light was seen to be composed of waves, the properties of which are different from
those of particles. Each theory was able to explain the behaviour of light in its own way. Each
also predicted the behaviour of light under given circumstances, and formulated experiments
that they could point to as exemplars. The difficulty that arose was that each paradigm was
able to validate its view for the behaviour of light with respect to its specific experiments, but
neither to the exclusion of the other. An eventual result was that a new paradigm of quantum
physics arose that regarded light as being able to manifest the properties of both corpuscles
and waves.
3
Systems thinking too has been changing, and indeed passing through its own phases across
the decades. It can be argued that prior to the 1970s systems operated under a single paradigm
[Jackson, 1992, p5]. However, new influences were underway that might today be connected
to the developing ideas of complexity. Action Research had been gathering support. It was a
development of the work of Gestalt-Field theorists who believed that successful change
requires a process of learning [Burnes, 1992, p166]. “It originated from a desire to alter and
improve social situations, or to help people in need. Its aim is to not only collect information
and arrive at a better understanding, but to do something practical as well. Sometimes, the
exponents of action research are dubious about the possibility of making detached and
scientific studies of human affairs. They may argue, for example, that an investigator cannot
but influence the behaviour of people he is studying, that experimentation is extremely
difficult, if not impossible, in the social sciences, that there is the intermediary of the human
instrument in measurement, and that all these vitiate the scientific status of social research”
[Mitchell, 1969, p2]. Argyle [1957] argues that action research should:

(a) prove that interventional activity is genuinely effective in making change,

(b) it should show the precise conditions under which interventions can result in desirable
outcomes.

A further development questioned whether systems thinking could deal with ill-structured and
strategic problems. To address this question, soft systems thinking and organisational
cybernetics arose [Jackson, 1992, p5]. The paradigmatic basis of the traditional approach
adopted a truth system that conflicted with those of the others. For instance, in soft systems
thinking the approach to inquiry centres on the weltanschauung principle that in addition is
concerned with the cultural attributes of stakeholders. In contrast, traditional “hard” systems
thinking ignores the idea of subjectivity, often by subsuming it within a pattern of behaviour
that the situation is perceived to be constrained by. The other approach, organisational
cybernetics, is specifically intended to deal with complexity by seeing a purposeful activity
system in terms of a dynamic relationship with a metasystem (see Beer [1979]) that controls
it. This provides a more macroscopic view of the situation, and shifts the focus from the
details of the complexity. Some critics of organisational cybernetics regard it as a hard
approach to inquiry, while others see it as soft. This is because it is an approach that is very
much inquirer determined, and may thus be operated according to a virtual paradigm
determined by the inquirer.

The ideas of Ackoff about systemic emergence are currently used as a way of solving the
problems of complexity and chaos. Chaotic behaviour can be dispelled, if we can find
“whole” or macroscopic emergent properties or patterns of behaviour.

3. Mapping Situations to Modelling Approaches

In order to be able to distinguish between different inquiry approaches and their ability to
handle situations, Harry [1994, p.255] created a two dimensional space. The purpose was to
map out the relationships between a situation and a modelling approach being adopted. He
introduced the two variables softness and structure:

 softness relates to the involvement of people and their mental perspectives,

4
 structure relates to the relationship between components of a model.

This space is shown in figure 2, where the vertical axis represents the soft/hard dimension of
a situation being modelled, and the horizontal axis of well/ill structure relates to the
modelling approach being adopted. An examples of how to interpret plots in this space will
be given now.

The selection of paradigms and their associated methodologies can be related to the situation
being examined. As an introduction to this, it is possible to show the relationships between
methodologies and problem situations simply by examining different hypothetical
combinations and seeing how these have in the past been used.

Consider the four points A, B, C, and D mapped in figure 2. Its interpretation can be found in
Harry [1987, p.256], assuming the following interpretation.

Situation

Soft A B

Hard C D
Methodology
Unstructured Well structured

Figure 2
An Approach to Map the Relationship Between Situations and Problems

Position A represents the situation where an unstructured approach is applied to a soft

situation. An example of this might be found where dispute occurs about the nature of a
situation, and people centred solutions are explored. How such a situation is solved is not
predeterminable.

Position B represents a situation where a structured approach is applied to a soft problem.

Here one attempts to deal with disputes using approaches like Soft Systems Methodology or
Organisational Development.

Position C represents an unstructured approach applied to a hard problem. This occurs for
example when the problem is clearly defined, and has objectively measurable criteria for
success. Such a situation is represented by the use of prototyping applied to the building of a
database system.

Position D represents a structured approach being applied to a hard problem. Thus, the
System Development Life Cycle or SSADM [Harry, 1994] are examples of methodologies
that can be applied to a situation which is apparently very well known .

4. Creating a Modelling Space

5
Like others in the post 1970s period, Rosenhead [1989] has been concerned with complex
situations, and in particular with the development of Operational Research systems
methodologies that can be used for complex situations. In pursuing this interest, he identified
three characteristics of complexity:

(a) that situations are more complex when they involve people;
(b) that complex situations may not be well-structured, in particular because cause-effect
relationships may not be determinable;
(c) that complexity is enhanced when situations are uncertain.

Let us explore this a little further to distinguish between each characteristic. With respect to
(a), when situations are considered in terms of people and their subjectivities, the view of the
situation is said be soft. On the other hand if people are seen as objects that are to be
manipulated, then the view is said to be hard. Considering (b), if a situation is seen to be well-
structured, then the parts that are seen to make it up (and their interrelationships) are well
defined across space or time. If this is not the case, the situation is said to be ill-structured.
Finally in (c), situations are seen on a scale of certainty to uncertainty that relate to the degree
of knowledge about them. A consequence is this relates to the predictability about the future
states of a given situation.

It is feasible to extend the map proposed by Harry to include Rosenhead’s ideas. To do this
we shall take hardness, structure, and uncertainty as three dimensions of consideration that
are analytically and empirically independent, and establishable as a frame of reference that is
indicative of the complexity of a situation. We shall refer to these dimensions as
orthogonalities, and the space that is define by them is called a modelling space [Yolles,
1996]. All systems paradigms should be susceptible to description through these
orthogonalities, and their position in a modelling space will be indicative of how much
complexity they are able to deal with. Our task is now to better describe these three
dimensions to illustrate their independence.

Hardness

This is related to the possible way the elements of a situation are seen. In entities that are
classed as hard, tangible things tend to dominate: that is, they are definite and examinable.
Their properties can be objectively defined and measured or assessed in some way that does
not depend on personal values. Soft entities on the other hand are relative to people and their
mental perspective. They have properties that cannot be measured objectively. Personal
values, opinions, tastes, ethical views, emotions, or weltanschauung are examples. People and
their psychological needs dominate. Softness is therefore directly related to subjective
mentality.

Whether a situation is classed as hard or soft will depend upon the view that is taken, and this
derives from the weltanschauung of an inquirer and the paradigm that has been adopted.
There are grades of hardness to softness, and these are normally seen to occur on a continuum
that we say passes through relatively hard/soft.

6
It has been suggested by a reader that the idea of softness and hardness can be defined in
terms of the degree of “ease with which agreement about the nature of a situation can be
achieved”. While an interesting idea, for us, this relates softness to the notion of conflict, that
occurs [Yolles, 1997] when individual subjectivities are involved and generate manifest
contesting differences that are world view originating. Here, if a situation is hard, then the
differences can be decontested easily. In a soft situation, the ability to decontest differences is
more difficult, and this can in addition be connected to emotionality. It may be worthy to
note, however, that ease or difficulty to decontest differences is a function of interest (in the
Habermas [1970] sense of the word). Thus when contesting differences become manifest, one
is forced to examine the individual subjectivities if one has an interest in agreement or
resolution, and this demands a soft approach. One cannot view a situation in this case from a
hard perspective unless agreement or resolution has been achieved. It would seem to be clear,
then, that relating softness to ease of decontesting differences offers a secondary rather than a
primary definition, and while useful is therefore not satisfactory.

Structure

This is related to the possibility of interrelationship among the elements of a situation.

Structure is about the relationships between definable entities like roles, objects or processes.
In ill-structured situations, the entities and their relationships are not well defined, whereas in
well-structured ones they are. Dynamic well-structured situations link entities across time in
causal relationships. As with softness, this is conceived as a continuum [Langley et al., 1987,
p.15] which may be qualitatively divided. The simplest qualitative division is to define the
term “semistructure” that lies somewhere between the poles.

A semistructured situation exists when neither a highly-structured nor unstructured situation

is found. Thus, a decision-making process involving well-known information about a
manufacturing process and unpredetermined ideas about where the process should be
directed, would be semistructured. It may be noted that the concept of semistructured
processes is important to the field of Decision Support Systems [Keen, Scott Morton, 1976;
Alter, 1980].There are variations on the term semistructured. For instance O’Brian [1995,
p396] proposes the term “moderately structured” situations. However, for our purposes we
would prefer to refer to partially structured or semistructured situations.

Structured situations may appear to be unstructured, if they involve entities which have
unpredeclared or invisible mutual relationships. We can thus distinguish between visible and
invisible structure, a concept which is relative to the individual who is looking. We can
alternatively talk of this by referring to deep (or cognitive) and surface (or behavioural)
structure. Situations that appear to be unstructured at a behavioural level when examined
more closely or in a different way, may be seen to have a conceptual relationship that is
defined at a cognitive level and has not been manifested. This is referred to as deep structure
[Keen and Scott Morton, 1976, p93; Chomsky, 1975]. This idea also relates to the concept of
relativity in that whether a situation is perceived to be well-structured is determined by the
context from which it is viewed. The reason for this is that if a group of entities appears to
have no connection one with another, this does not mean that they do not have a group
coherence that gives meaning to the entities as a group. A manifestation of that meaning can
occur at the surface level, and can be transformed to a purpose or set of purposes for the
entities of the group that become manifest behaviourally. Thus, it is at the surface level,
7
where the entities operate to carry out their functions or purpose, and the connection may be
invisible. Whether a deep structure exists or not will depend upon how an inquirer sees, and
the paradigm they use in order to do the seeing.

Uncertainty

This is spatially related to the possible knowlege available about a situation, or over time to
any possible outcomes that derive from actions. For instance, certainty occurs when we know
that each choice of action is linked with only one particular outcome. Uncertainty occurs
where there is a plurality of possible outcomes resulting from one of many choices of action.
We do not know which will result from a given action, and in any case we cannot assign
probabilities to them, or even identify possibilities for them. It also relates to the technical
nature of the situation, a term adopted by Habermas [1970] to refer to the control aspects of a
situation and its future states or predictibility.

We can therefore conceive of a certainty-uncertainty continuum defining an axis of

variability, and we can differentiate between them with an intermediate graduation of
relatively certain or relatively uncertain.

The Modelling Space

In short therefore the three orthogonalities that define complexity are:

 softness that relates to the subjective mentality that is attributed to a situation, and this can
vary with the degree of subjectivity importance to the world view of an inquirer
 structure that relates to perceivable relationships between arbitrarily definable entities
within a situation, and this can vary with the richness of its interrelationships that an
inquirer sees; it is thus world view dependent.
 uncertainty about the nature of a situation that will vary with the knowledge about it, is
connected to world view, and relates directly to future outcomes and predictability.

To represent an event (say a view about a problem situation) in this modelling space we
establish the set of coordinates (certainty, softness, structure) that defines a position as shown
for instance in figure 3. The space is a bounded cube with sides have been normalised to vary
between a measurement of 0 and 1. These units are not intended to be indicative of a precise
measurement scale, but are manifested from a qualitative evaluation that translates to a fuzzy
point somewhere between these values.

Certainty

1 Structure
PS
1

0 1 Hardness

8
 Figure 3:Example of a Modelling Space for a Problem Situation (PS)

5. Changing Paradigms to Accommodate Complexity

We have said that paradigms change as they mature and other paradigms sometimes come to
replace them. Our interest here is to illustrate changes in Operational Research paradigms.

5.1 The Traditional Operational Research Paradigm

Rosenhead [1989] discusses the recent history of Operational Research in terms of its
changing paradigm. In its traditional light, a view of Operational Research is that it is a
modelling process for problem solving that consists of the five steps shown in table 1.

1. identify objectives with weights,

2. identify alternative courses of action,
3. predict consequences of actions in terms of objectives, usually as a cause-effect relationship,
4. evaluate the consequences on a common scale of value,
5. select the alternative whose net benefit is highest, that is the optimal solution.
Table 1
Traditional Paradigm in Operational Research

This approach was used for many years, until it was realised that while attractive because it
created models of problems that could be solved, the solutions did not correspond with reality
except in very special cases. Difficulties that many assigned to the traditional approach lay in
the idea that it:

1. was deterministic, which meant problems were assumed to be certain

2. did not consider people as having needs, so that problems were assumed to be hard
3. assumed modelling relationships between entities in a situation were known, supposing
that problems were well-structured.

5.2 The Dominant Operational Research Paradigm

Determinism was shown to be inadequate in modelling situations when it was realised that
the models were frequently far from complete explanations, and solutions were interesting
rather than useful. Certainty was an assumption that was untenable in a world that seemed to
be uncertain. One answer lay in a new approach through the application of Baysian statistics.
Since it was seen that futures could not be foretold, the idea arose that probabilities could be
used to generate future expectation. In this paradigm shift that Ackoff [1979] refers to as
“predict and prepare”, existing certainties are replaced by probability estimations, and these
are then assumed to be valid for future situations. Thus, certainty was replaced by relative
certainty. With the addition of statistical theory, the new paradigm still maintained the
traditional set of propositions. One of the difficulties with this view was that the modelling of
futures in which the probabilities changed was not permitted, and it was therefore assumed
that this did not happen. Rosenhead refers to this view as the dominant paradigm of
Operational Research, the assumptions of which are identified in table 2.

9
1. Problem formulation occurs in terms of a single objective that is optimisable; there may be multiple
objectives that, if recognised, may be traded off one against the other on some form of common scale.
2. There are overwhelming data demands, with accompanying problems of data: distortion, availability,
and credibility.
3. Consensus is assumed possible, with the approach adopted assuming depoliticisation, and
scientificisation.
4. People are treated as passive rather than active participants in the situation.
5. There is an assumption of a single decision maker with abstract objectives from which concrete actions
can be deduced for implementation through a hierarchical chain of command
6. Attempts are made to abolish future uncertainty, and pre-take future decisions.
Table 2
Dominant Paradigm for Operational Research

5.3 The Need for a New Paradigm in Operational Research

The difficulty with the dominant paradigm as discussed by Rosenhead still lay in its
adherence to problems that were assumed to be hard and well-structured, even though relative
certainty was now a feature. For our purposes, explanation can be presented best by
considering each of these dimensions in turn.

Firstly, let us consider uncertainty. Data is needed to identify what is happening in situations,
and this may be seen to be wrong or incomplete. There may be a problem with the
relationship between data collection and problem location, and some dimensions of
consideration will have intangible elements that are not quantifiable. Confidence over data
collection may therefore be inappropriate. Without adequate information about a situation, it
is not possible to formulate conclusions about the parts that make it up and the problems it
has. Additionally, it is not possible for any objectives to be sensibly taken up in a course of
action, nor the easy identification of the probable consequences that might develop.

Uncertainty is an important consideration when evaluating situations and the way in which
decision makers make decisions about them. Hopwood [1980] offers a typology from which
we produce table 3. This relates uncertainty to course of action in connection to both goals
and consequences. Thus, the greater the uncertainty, the more decision makers tend to rely on
soft (human mentality and values) approaches rather than information in relating objectives
for action with anticipated consequences.

Consequences of action Goals for Action

Certainty Uncertainty
Certainty deterministic/probabilistic decisions decisions under bargaining
Uncertainty judgmental decisions decisions through inspiration
Table 3
Typology for decision making under uncertainty, relating objectives for courses of action to
consequences

When considering the dimension of hardness, we see that people are involved in situations
and directly influence how they change. As Rosenhead argues: “an organisation is not an
individual...Decisions and actions emerge out of interactions between a variety of actors
internal to the organisation. Each may indeed have an individual perspective or world view
(weltanschauung) through which actions or statements of others are interpreted. What the
constraints are, what the priorities should be, what the problem actually is, may be perceived
10
quite differently....A process of accommodation is necessary before a problem can emerge
which can carry assent and commitment to consequential actions” [Rosenhead, 1989, p9].

The idea that participants to a situation are purposeful, and that decisions are a consequence
of group processes in which conflicts sometimes have to be resolved, is a feature of
modelling that should be addressed. This leads to the idea that different situations will be
distinct from one another because they are composed of different groups of people.
Situational uniqueness is therefore a consequence of softness, and this perspective abolishes a
hard approach.

Having said this, it is appropriate to note that in organisational situations, we are normally
concerned with situations involving small groups of people, hence the use of soft
methodologies. In the social science literature where the modelling of these groups comes
from, there is a difference between the interactive processes of large and small groups, the
latter tending to be less predictable. In larger group situations, patterns of behaviour can
develop, be recognised, and sometimes be predicted with some degree of success. One
example of this is represented by Hitler’s ability to predict the response of crowds and control
them during the build up to the second world war. Another is the idea that the population’s
voting behaviour can be predicted because it is such a large scale phenomenon. However, this
is difficult to do because of the complexity of issues that can affect the value judgements of
people. Situations involving large groups may therefore be considered to be less
individualistic than small groups. This occurs when groups achieve what is referred to as
“critical mass”, a term analogous to nuclear processes. It suggests that groups, when having
reached a particular threshold of size and therefore complexity, establish formalised patterns
for classes of behaviour which to some extent and under certain reasonable conditions are
predictable. Such situations may be considered to be classified as relatively soft.

To consider now the last dimension of interest, problems may not be well-structured. To
distinguish between well and ill-structured problems, it is useful to introduce the term unitary
and pluralistic:

 A unitary situation consists of a set of identifiable parts that have a unique purpose
and single set of objectives.
 A pluralistic situation occurs when there exists a set of parts that (a) represent
different aspects that are not clearly definable, or (b) which have purposes that may be
incommensurable and thus in conflict.

Pluralistic situations have goals that cannot be easily assigned to the parts in such a way that
they do not clash. Modelling approaches that do not take account of pluralistic situations
cannot work because courses of action cannot be defined for situations that are unclear. Such
situations may thus be semistructured or unstructured, with some parts that cannot be clearly
related one with another if indeed all parts are known.

In the event that parts to a situation are themselves well-structured, their relationship to other
parts may not be at all well-structured. Well-structured problems are not only normally
assumed to be unitary, but also to have firm constraints, and establishable time related
relationships between cause and effect.

11
5.4 The Rosenhead Paradigm

The approach of the dominant paradigm in Operational Research contrasts with the thinking
of Rosenhead, whose paradigm is sensitive to the needs of Operational Research as explained
above, and is offered in table 4. As a consequence, the modelling techniques and
methodologies proposed by Rosenhead operate in a modelling space that is uncertain, soft,
and unstructured. While the dominant paradigm uses a calculus of probabilities, Rosenhead
seeks rather a calculus of possibilities, and reflect more complexity in situations. This is a
requirement that presupposes unstructured or semistructured situations under uncertainty or
relative uncertainty using methodologies that enable options to be defined and explored.

1. Non-optimising, looking for alternative solutions acceptable on separate dimensions, without trade-offs.
2. Reduced data demands, achieved by greater integration of hard and soft data with social judgements.
3. Simplicity and transparency, aimed at clarifying the terms of conflict.
4. Conceptualise people as active subjects.
5. Facilitating planning from bottom up.
6. Accepts uncertainty, and aims to keep options open for later resolution.
Table 4
Rosenhead Complexity Paradigm for Operational Research

6. Logging Changes in the Operational Research Paradigm

It is possible to show that paradigms and the perception of situations do change in time
graphically by attempting to estimate qualitative movements in quantitative terms. One way
of doing this is shown here.

The appearance of a new paradigm must be able to be differentiated from an earlier paradigm
in the modelling space. To do this for ease of modelling and comprehension we need to create
an aggregate value that we propose to derive from the three dimensions of uncertainty,
hardness, and structure. If we can do this, then since the aggregate will represent the degree of
involvement of each of the three characteristic variables, the resultant value will be an
indicator of how well the paradigm is able to deal with complexity.

If we are able to find values that can be assigned to each paradigm for these three dimensions,
then the aggregate value can be determined using a technique of numerical analysis referred
to as the Euclidean norm [Wilkinson, 1965]. This is equivalent to generating a mean vector in
the modelling space of the movement, and taking its absolute size to be between (0,1). In
doing this, the aggregate is obtained by squaring each term, and summing the result. This
must be normalised to restrict it to its bounds, and this occurs by dividing by the maximum
sum of the squares to bound the result. When plotted against time (decades), it should show
how new paradigms are able to cope with complexity.

The first requirement in doing this lies in plotting paradigm positions in a fuzzy region as they
occur in the modelling space. As a subject of this exercise, we have choosen the Operational
Research paradigm.

When we assign quantitative co-ordinate values to paradigm positions in the modelling space,
they must be seen as representative of qualitative plateaus. They demonstrate a technique of
assigning quantitative values to qualities typical of the approach taken in the domain of
12
Artificial Intelligence to represent qualitative human thinking. In the table 5 below we offer
landmark values (see for instance Kuipers [1986]) that are intended to represent different
qualitative descriptions through the creation of regions that we represent by a single landmark
altitude.

Qualitative Description Landmark Values

Certain , hard, well-structured 1
Relatively certain/uncertain, relatively 0.5
soft/hard, semistructured
Uncertain, soft, ill-structured 0
Table 5
Assigning qualitative properties to regional landmark altitudes.

The traditional Operational Research paradigm is located in the modelling space with a co-
ordinate landmark (certain, hard, structure) of (1,1,1), that is the paradigm operates with
situations that are certain, hard, and well-structured.

Some years after its use, a paradigm change occurred, and certainty was replaced by
probability to be better able to predict events. We shall say that certainty was replaced by
“relative certainty”. Thus, the modelling space co-ordinate (certainty, harness, structure)
becomes (1,0.5,1).

The Rosenhead paradigm shows a shift to give a new modelling space co-ordinate landmark
vector in (certainty, hardness, structure) of (0,0,0).

For our purposes it would be useful to be able to identify at least one other paradigm. To do
this we will interpolate, thus supposing the appearance of a paradigm not normally discussed
in the Operational Research literature. Prior to the Rosenhead paradigm a method existed that
could deal with relative softness and semistructure, like the modelling technique of Fraser and
Hipel [1984] called Conflict Analysis. The new co-ordinates of (certainty, hardness, structure)
are (0,0.5,1). It is supposed that (a) we are totally uncertain about the outcome of a conflict,
(b) that organisations are involved and their paradigms must be taken into consideration
whilst still trying to address the situation, and (c) that it is known who the participants to the
conflict are, and what their relationship is - that is the situation is highly structured.

Four Operational Research paradigms are represented:

(1) the traditional paradigm which assumes situations to be certain and therefore purely
deterministic, hard, and well-structured
(2) the dominant paradigm which appends to the idea of certainty (that is determinism) that of
probability
(3) an interpolated Fraser and Hipel paradigm, which also supposes semistructure and the
possibility of influence by groups of people (relatively soft)
(4) the Rosenhead paradigm that supposes uncertainty, softness, and ill-structure.

We can calculate aggregates for each paradigm coordinate in order to generate a mean value.
This can be plotted across the decades to indicate that indeed, with respect to the generic
characteristics of the modelling space, the paradigms do indicate movement. This has been
done in table 6 by adopting the Euclidean norm (the normalised sum of the squares of the
13
coordinate values). The aggregate values are plotted in a modelling space in figure 4 that is
intended to illustrate how the Operational Research paradigms have moved over a period of
say 4 decades. They are intended to illustrate the degree of complexity that a given paradigm
is able to cope with. Highest levels of complexity occur when the aggregate value is at a point
0, while lowest levels occur at an aggregate value of 1.

Type of Date Qualitative position of Euclidean Coordinate first First

Paradigm period paradigm aggregate difference difference
(certainty, hardness, aggregate
structure)
Traditional 1940’s (1,1,1) 1
(0,-0.5,0) 0.08
Dominant 1 1950’s (1,0.5,1) 0.75
(-1,0,0) 0.30
Fraser & Hipel 1960’s (0,0.5,1) 0.42
(0,-0.5,-1) 0.42
Rosenhead 1980’s (0,0,0) 0
Table 6
Calculating Modelling Space Aggregate Values for Operational Research Paradigms

Paradigm aggregate value

indicating ability to handle
complexity

simple 1 

0.75 

0.5


0.25

complex 0 

1940’s 1950’s 1960’s 1980’s time t 

Figure 4
Appearance of new Operational Research paradigm over the decades

We are also able to calculate first differences of the paradigm positions that are indicative of
how the paradigms change. The aggregates generated are indicative of the degree of change in
dealing with simple and complex modelling processes. The result is shown in figure 5.

14
 Change in the way paradigms
can deal with complexity
0.5

0.4

0.3 

0.2

0.1

0 time (decades)
1 2 3 4
Figure 5
Aggregate Changes of Paradigm in Modelling space over the decades

7. Conclusion

Paradigms have a capacity to change, but they may be bounded by the very conceptualisations
that at one time made them successful. The developing paradigm of complexity has emerged
over the last few decades, to enable us to explain phenomena that traditional paradigms are
unable to explain. In management systems for example, the traditional paradigm has given
way to other paradigms, not because it is incapable of describing events in terms of
emergence, but because the models that it generates may not always identify the emergent
properties or patterns in a way that is seen (at least by some) to be appropriate. Other
paradigms, like that of managerial cybernetics or soft systems thinking have provided, for
some people, a more useful approach to deal with complexity. Soft systems thinking
conceptualises that people and their subjectivities are important to situations. The
involvement of the participants in a situation will offer a variety of views that will (hopefully)
deal with complexity. By characterising the degree of complexity of a paradigmatic view in
terms of structure, hardness, and uncertainty, we attempt to illustrate that indeed Operational
Research paradigms follow a virtually straight pathway from simplicity to complexity when
measured against these variables.

8. References

Alter, S.L., 1980, Decision Support Systems: Current Practices and Continuing Challanges.
Addison-Wesley, Reading Mass., USA
Ackoff , R.L., 1979, The Future of Operational Research in the Past, J.Opl Res. Soc., 30,93-
104.
Argyle, M., 1957, The Scientific Study of Social Behaviour,
Berger, P., Luckman, T., 1966. The Social Construction of Reality. Penguin
Burnes, B., 1992, Managing Change. Pitman Publishing, London.
Checkland, P.B., Scholes,J., 1990, Soft Systems Methodology in Action. John Wiley & Son,
Chichester
Chomsky, N., 1975, Problems of Knowledge and Freedom. Pantheon, New York
Fraser, N.M., Hipel, K.W., 1984, Conflict Analysis, Models and Resolutions. North Holland
Guastello, S.J., 1997, Science Evolves: An Introduction to Nonlinear Dynamics, Psychology,
and Life Sciences. Nonlinear Dynamics, Psychology, and Life Sciences. 1(1)1-6.
Habermas, J., 1970, Knowledge and interest. Sociological Theory and Philisophical Analysis,
pp36-54, (Emmet, D., MacIntyre, A., eds), MacMillan, London
15
Harry, M., 1994, Information Systems in Business. Pitman, London
Hoffman, B., 1947, The Strange Story of the Quantum. Penguin books, Middlesex, UK
Hopwood, A.G., 1980, The Organisational Behavioural Aspects of Budgeting and Control, in
Arnold, J., Carsberg, B., Scapens, R (Eds.), Topics in Management Accounting,
(pp221-40, Philip Allen, Deddington
Jackson, M.C., 1992, Systems Methodologies for the Management Sciences. Plenum, New
York.
Keen, P.G.W., Scott Morton, M.S., 1978, Decision Support Systems: an organisational
perspective. Addison-Wesley
Kuhn, S.T., 1970, The Structure of Scientific Revolutions.University of Chicago Press,
Chicago
Kuipers, B., 1986, Qualitative Simulation. Artificial Intelligence, vol. 29.
Kyberg Jr., H.E., 1968, Philosophy of Science: a Formal Approach. Collier-Macmillan,
London.
Langley, P., Simon, H.A., Bradshaw, G.L., Zytkow, J.M., 1987, Scientific Discovery. MIT
Press, Massachusetts, USA
Mitchel, G.D., 1968, A Dictionary of Sociology. Routledge and Kegan Paul, London.
O’Brian, B., 1975, Information Management Decisions: Briefing and Critical Thinking.
Pitman Publishing, London
Patching, D., 1990, Practical Soft Systems Analysis. Pitman Publishing
Rosenhead, J., 1989, Rational Analysis for a Problematic World. John Wiley and Son
Sorokin, P., 1964, Comments on Schneider’s Observations and Critisms. In Zollschan, K.G.,
Hirsch, W., (eds), Explorations in Social Change, Routledge & Kegan Paul
Wilkinson, J.H., 1965, The Algenraic Eigenvalue Problem. Oxford University Press
Yolles, M.I., 1996, Critical Systems Thinking, Paradigms, and the Modelling space, Systems
Practice, 9(3).
Yolles, M.I., 1997, Managing Complexity: a Viable Approach to Inquiry. Under completion.

16